Model is adding extra words/phrases

[Dias2019]

While testing this model, I noticed several cases when the model adds extra phrases or words as shown in below example. Tbh, this makes the model quite risky for usage due to its unpredictable corrections like this one. Therefore, I was wondering if others are facing the same issue? or Am I making some mistake during installation/running of the scripts?

My input (with misspellings):

Background: I was 10 loking to hire a junior dfata scientist to join my datsa sciene team. 
I wan tod that 150 here's a great skrll shortage , especially in te IT and data fields .'

The output of a model:

Background # I was 10 looking to hire a junior data scientist to join my data science team. 
I want to hire a junior data science team .' or that 150 here\'s a great skrll shortage , especially 
in the IT and data fields .''

As you may notice, the phrase "to hire a junior data science team" was added in the second sentence, even though it wasn't supposed to do that.